US7756320B2 - Defect classification using a logical equation for high stage classification - Google Patents
Defect classification using a logical equation for high stage classification Download PDFInfo
- Publication number
- US7756320B2 US7756320B2 US10/794,267 US79426704A US7756320B2 US 7756320 B2 US7756320 B2 US 7756320B2 US 79426704 A US79426704 A US 79426704A US 7756320 B2 US7756320 B2 US 7756320B2
- Authority
- US
- United States
- Prior art keywords
- classification
- defect
- defects
- criteria
- classification criteria
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active, expires
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30148—Semiconductor; IC; Wafer
Definitions
- the present invention relates to a defect classification method applied to industrial products utilizing a thin film technology and, more specifically, to a method for automatically setting classification criteria for semiconductor products with which detailed defect review and classification are considered important after defect detection through post-step inspection in the automatic defect classification method.
- ADC Automatic Defect Classification
- Such ADC is varied in type, and so far proposed are: the rule type based on a rule predetermined for classifying, into defect classes, defect features containing a plurality of image features extracted from images such as image brightness and defect shape; the teaching type for automatically creating criteria for defect classification based on defect class distribution in a multi-dimensional vector space by regarding defect feature items each being a scalar value as multi-dimensional vectors; and the combination of the rule type and the teaching type.
- the rule type generally requires setting of a determination threshold value each to various defect feature items, and the teaching type requires to derive defect class distribution in the multi-dimensional vector space.
- the defect classification criteria is set based on the defect feature distribution prior to automatic classification.
- the problem of the conventional technology is that the defect feature distribution may look different once automatic classification is started, and if so, automatic classification cannot be appropriately done. For example, if the defect samples prepared at the time of classification criteria setting do not include any important ones, no classification criteria will be set for the defect classes having no samples. Thus, automatic classification cannot be normally done if no measure is taken therefor.
- Prior to automatic classification it is unknown whether a target semiconductor layer has any defect different in type from the defect samples which have been already collected. Therefore, it is difficult to determine whether the classification criteria set at this point of time is good enough or not.
- teaching data is created for defect classification first by calculating the amount of features from defects detected from a semiconductor wafer to allocate the result to the feature amount space, and then by performing category assignment from the defect distribution in the feature amount space.
- the same process is applied to any defects detected for classification from the semiconductor wafer for comparison with the teaching data.
- the teaching data is to be accordingly corrected based on the observed difference.
- the issue here is that the recent process control requires defect classification in a detailed manner, and to make it a reality, the number of dimensions in the feature amount space is getting increased. Therefore, it is difficult to automatically detect the change of the feature amount distribution because the defect distribution is initially derived from the small number of samples. Further, even if the classification criteria is determined as having a problem in its setting through defect monitoring after automatic classification is started, actually, the classification criteria has been considerably difficult to appropriately reset.
- An object of the present invention is to estimate the appropriateness of automatic defect classification, and to make classification criteria settable with a guarantee for the better classification performance. Another object thereof is to realize a resetting technique for the classification criteria with which the better classification performance is guaranteed in its entirety.
- one typical aspect of the present invention is directed to a defect classification method, which includes: a step of calculating classification criteria based on a plurality of teaching defect groups; a step of classifying a defect cluster known with a true classification class using the corresponding classification criteria; a step of calculating an accuracy ratio of a case where the classification criteria is used based on a difference between a classification result using the classification criteria and the true classification class; a step of calculating the relationship between the classification accuracy ratio and the number of teaching defects used for calculation of the classification criteria; a step of calculating a classification criteria A from defects known with their true classification classes; a step of estimating the classification accuracy ratio with respect to the defects known with their classification classes with the classification criteria A by referring to the relationship between the number of teaching defects and the classification accuracy ratio; and when the estimated classification ratio is high, a step of determining as automatic defect classification being executable using the classification criteria A.
- FIG. 1 is a diagram illustrating the basic sequence of a defect classification method in one embodiment of the present invention
- FIG. 2 is a diagram illustrating the classification criteria setting sequence in the defect classification method in the embodiment of the present invention
- FIG. 3 is a diagram illustrating the basic characteristics of the classification criteria to be evaluated when the classification criteria setting sequence of FIG. 2 is executed;
- FIG. 4 is a diagram illustrating the classification criteria change sequence in the defect classification method in the embodiment of the present invention.
- FIG. 5 is a diagram illustrating the sequence for selecting a defect to be used as a reference for classification criteria setting in the defect classification method in the embodiment of the present invention
- FIG. 6 is a diagram illustrating an exemplary teaching screen of the classification criteria in the defect classification method in the embodiment of the present invention.
- FIG. 7 is a diagram illustrating an exemplary case where the method of the present invention is incorporated into a review SEM.
- FIG. 8 is a diagram illustrating a classification criteria resetting timing in the defect classification method in the embodiment of the present invention.
- FIGS. 1 to 8 an embodiment of the present invention is described by referring to FIGS. 1 to 8 .
- the pattern of which is formed to be of multilayer structure by going through a plurality of steps.
- every layer is subjected to outer appearance inspection, and if any defect is detected thereby, the defect is reviewed and then classified according to defect type.
- various manners have been proposed and put into practical use as follows:
- FIG. 1 shows the classification sequence of ADC of the present invention.
- Step 101 is a classification task step using a manual, and therein, defect images that have been plurally collected are classified through visual evaluation.
- Step 102 is a step of setting classification criteria using the defects having been completed with visual evaluation in Step 101 . This Step 102 is described in more detail by referring to FIG. 2 .
- Step 10 when a determination is made that the defects having been subjected to manual classification is small in number, defect images are plurally collected again for repeating Step 101 .
- Step 103 is an image acquisition step, in which any defect images are to be acquired for classification using the classification criteria set in Step 102 .
- An acquisition method applicable thereto may be the one for transferring any image data that is already at hand to a memory region for executing a program by communications means and others, or the one for picking up any new images by such a detection system as shown in FIG. 7 , which will be described later.
- Step 104 is an automatic classification step, in which classification is applied to any defects unknown with their classification results.
- Step 105 is a check determination step, in which a determination is made whether the classification criteria set in Step 102 requires evaluation or not. An embodiment of Step 105 will be described in detail by referring to FIG. 8 . If the determination made in Step 105 tells that the classification criteria requires no evaluation, another determination is made in Step 108 whether the classification sequence is through. If determined in Step 108 that the sequence is through, the classification sequence is accordingly ended. If determined that the sequence is not yet through, the procedure returns to Step 103 .
- Step 106 the procedure goes to Step 106 to go through an automatic classification result check step, in which the defects having been automatically classified in Step 104 are partially evaluated by using a manual to see whether automatic classification has been correctly done in Step 104 . If evaluated in this step as automatic classification having been correctly done, the procedure returns to Step 103 for picking up any images unknown with their classification result, and then Step 104 for automatic classification of the images. If the check result in Step 106 tells that the ratio of the correct automatic classification result is low in accuracy ratio, the classification criteria is set again in Step 107 using the images that have been evaluated using the manual in Step 105 , and then executes the sequence from image acquisition in Step 103 to automatic classification result check in Step 106 .
- the typical defect features are image features and substance composition features.
- image feature extraction so far proposed is the one utilizing defect regions to calculate defect size, shape feature, defect brightness, image texture, and surface unevenness.
- the defect regions are those calculated through comparison of two images: one is the image including a defect; and the other is the image of a region including no defect but having the same pattern as the other image.
- substance composition features exemplified is spectrum derived by Energy Dispersive X-ray Fluorescence Spectrometer (EDX), in which defects are exposed to electron beams and the resulting X-rays are analyzed.
- a threshold value is needed for use as a distinguishing factor. For example, such a feature as unevenness degree being the total sum of the shading level in the entire defect region is calculated, and if the result is exceeding a predetermined threshold value, it is determined as uneven, and if not, determined as flat. In such a manner, classification criteria is set. For setting of the threshold value indicating such a classification boundary, defects known with their attributes are collected in advance for achieving classification in the most appropriate manner.
- attributes in this example are those being flat, a bump, or a dip.
- the feature amount space is formed for teaching based on the amount of features extracted from defect images. Described therein is the method for calculating a discriminant that is used for discriminating teaching points.
- the teaching points are points in the feature amount space represented by feature amount vectors having been extracted from defect images to be used for calculation of the feature space.
- the feature amount vectors are n-piece combinations (c 0 , c 1 , . . . , cn) of feature amounts extracted from one image, where c denotes the feature amount extracted from one defect image, and n denotes the predetermined number of feature amounts.
- defect collection is generally done for use as a basis thereof.
- classification criteria setting is accordingly done.
- a number of defects are required to be collected for correct automatic classification.
- no such classification criteria is settable no matter how many images are collected. Therefore, if the classification performance fails to reach its sufficient level after classification criteria is once set, with no idea whether the layer is incapable for automatic classification or image collection is not enough, there is no choice but to again go through image collection and resetting of the classification criteria.
- FIG. 2 shows a sequence for classification criteria setting of the present invention.
- a value N is set for dividing the defect clusters collected in Step 101 into N groups, where N is a certain number.
- a value 2 is considered convenient for the initial value of N.
- Step 202 defects in the defect clusters are allocated in such a manner that each of N groups includes the defect.
- Each defect is provided with its true classification class by a manual in Step 101 , and allocation may be so performed that each group includes each defect type in equal amount.
- Step 203 one out of N groups is selected as a classification group, and the remaining N ⁇ 1 group(s) are selected as teaching groups. Then, the classification criteria is set by referring to the defects included in the teaching group(s).
- Step 204 the defects belonging to the classification group are classified according to thus set classification criteria. Because of N groups, there are N different possibilities of selecting one group as a classification group. Thus, Steps 203 and 204 are repeated until every possibility is through. Once N different possibilities are through, derived are the classification results for all of the defects based on the classification criteria, which is obtained from the teaching sets including no classifying defects.
- Step 201 the value of N is increased for another process. When the number of N is small, the process may be executed until the number reaches the defecting number. When the number of N is large, the process may be executed until the number reaches about 10 to 20.
- a defect class i the defecting number used for classification criteria setting is presumably Ci, and referred to as teaching defects.
- derived is an accuracy ratio for automatic classification when the number of teaching defects changes from Ci/2 to Ci*19/20.
- the larger the teaching defects in number the better the accuracy ratio of classifying the defects not used for classification criteria setting becomes.
- derived is such a relationship between the number of teaching defects and the accuracy ratio as shown in FIG. 3 . From the classification accuracy ratio of a case where the teaching defects are N being the largest, the classification performance can be estimated for a case where every defect in Step 101 is taught. If this classification accuracy ratio is high enough, the accuracy ratio is considered high of a case where any arbitrary defects unknown with their classification results are automatically classified based on the calculated classification criteria.
- the curve of FIG. 3 may be used as a determination factor. Referring to FIG. 3 , when the differentiation value of the curve is positive with N being the largest, the performance can be expectantly improved by increasing the number of teaching defects. In this case, the cause is determined as the shortage of the teaching defects. If the differentiation value of the curve is closer to 0 with N being the largest, it means that the classification accuracy ratio is already saturated, impossible for better performance even if the number of teaching defects is increased. That is, no high performance is expected with ADC.
- Step 205 a graphic display is made for the number of teaching defects—classification accuracy ratio for each classification class and in terms of the comprehensive performance. Further, if the performance is not high enough, a report is displayed to show whether there requires to increase the teaching defects in number or no expectation for better performance. Based on thus displayed report, if the performance is not high enough, the teaching defects are increased in number to repeat Steps 201 to 204 , and then the evaluation result is displayed again in Step 205 . In such a manner, after evaluating the classification accuracy ratio, in Step 206 , classification criteria is generated by using all of the teaching defects. The resulting criteria is regarded as the classification criteria to be generated in Step 102 .
- Step 104 The teaching defects to be collected at the time of Step 102 are guaranteed to be classified with a high performance by the sequence of FIG. 2 .
- the problem here is that the performance cannot necessarily be continuously kept because the defect type may not be the same for a long period of time. Therefore, there needs to monitor defect classification using a manual to see if it is correctly performed for a certain period basis. For the purpose, it is desirable to incorporate a monitoring rule into the classification criteria for ADC.
- the monitoring rule determines the monitoring timing or defect cluster to monitor based on the following external inputs, i.e., based on at least one of
- the stably-occurring defects are already in the images collected in Step 101 at the time of classification criteria setting, and thus there needs to pay special attention to defects unexpectedly occurring.
- the manner of above 1 first the time-serial change of the defecting number observed in a wafer is subjected to analysis. If the defecting number is considerably a lot in the wafer, the classification result will be monitored. In this manner, the classification result of any defects rarely occurring under the state that the process is stably executed can be evaluated, and the evaluation is easily incorporated into the teaching defect cluster.
- the spatial distribution of the defect position in the manner of above 2 random defects and systematic defects can be derived from the distribution. Accordingly, in a case where the spatial distribution which has not been observed so far is observed, it becomes possible to see how the defects having different features from the teaching defects are classified through monitoring of the classification result.
- One main object for defect classification is to identify the cause of lowering the product yield.
- a method for identifying the defect cause popularly applied in the actual manufacturing lines is the electrical inspection and the defect composition analysis.
- any defects highly critical in electrical or physical sense are needed to be subjected to defect classification with a high accuracy.
- By checking classification result derived as such, realized is automatic classification capable of estimating the defect cause and yield with high accuracy.
- FIG. 8 shows such details.
- a reference numeral 801 shows a classification criteria check timing that has been previously determined when the classification criteria is set. This check timing will be set long if the defects used for setting the classification criteria in Step 102 are large in number, and if small in number, the check timing will be set short.
- the monitoring result of the above monitoring rule at least singly selected from (1) to (5) changes to a greater degree, the check timing is changed in accordance with the change, for example.
- a reference numeral 802 shows the time-series change of the number of the defects detected per wafer described in the monitoring rule (1). Reference numerals 803 and 804 both show the number of defects as a result of detection where any characteristic patterns are observed in the defect position distribution detected per wafer.
- 803 shows the number of defects as a result of detection where any linear patterns are observed in the defect position distribution
- 804 shows the number of defects as a result of detection where any annular patterns are observed in the defect position distribution
- a reference numeral 805 shows the time-varying number of chips determined as defects as a result of electrical text
- a reference numeral 806 shows the number of defects in which any iron elements are found after EDX analysis.
- the defects having been subjected to automatic classification in Step 104 are used to extract a defect cluster to be checked for their classification state, and the extraction result is displayed for the user.
- This process may be done by a review device for picking up defect images, or a computer terminal connected over a network.
- the user may provide the manual-classification result to the defect cluster thus extracted by the ADC system so that the Step 106 can be realized for automatic classification result check.
- the classification result derived in Step 106 is preferably used for resetting of the classification criteria as it is.
- the teaching defects used for initial setting and the defect cluster checked in Step 106 are all used as the teaching defects for generating the new classification criteria in Step 107 .
- the teaching defect cluster applied with the classification criteria through the sequence described by referring to FIG. 2 can be checked with its performance.
- the defects other than those having been subjected to classification in Step 104 and provided with the classification results by manual in step 106 are unknown with their true classification results.
- a lot of phenomena of lowered classification performance are so far reported due to performing unbalanced teachings.
- To generate the classification criteria in Step 107 required is the guarantee not causing the classification performance to be reduced.
- FIG. 4 shows a process to be executed in Step 107 of FIG. 1 .
- Step 401 is a step of setting classification criteria using any given teaching defects.
- Step 402 is a step of extracting any difference of the classification results derived by classifying the defects having been classified in Step 104 partially or entirely using both classification criteria before and after its change. By checking only the defects extracted in Step 402 as determined as having any difference from the classification result, the defects having been through comparison in Step 402 can be evaluated by performance before and after the change of the classification criteria.
- Step 403 only to the defects classified and determined as having any difference in Step 402 , their true classification results are taught by a manual. If the initial classification criteria generated in Step 102 is already adjusted sufficiently, under normal circumstances, the defects showing any difference in the classification result will be small in number. In comparison between the true classification result input in Step 403 and the classification result in Step 402 for evaluation, if the classification performance by the classification criteria generated in Step 107 is lowered than the initial classification criteria generated in Step 102 , the defects subjected to manual classification in Step 403 are again added to the teaching defect cluster, and the classification criteria is set again in Step 401 . By establishing such a sequence, provided is the classification criteria capable of guaranteeing the better performance through addition of the teaching defects.
- the teaching defect images are normally picked up. From some images, however, the defect regions cannot be correctly extracted due to the image pick-up condition such as charging of SEM images, or even if extracted, the result may be unusually bright. If the classification criteria is set by using the resulting defects, the feature amount that is not originally supposed to distribute is used for criteria resetting. As a result, the classification performance will be lowered.
- Step 501 a defect is designated from a teaching defect cluster as a determination factor.
- Step 502 the teaching defects except for the defect designated in Step 501 are divided into N groups.
- N denotes the number of teaching defects or smaller, and may be divided in advance to be about 20.
- such division is applied to equalize the defecting number included in each defect class in each group.
- step 503 one group is selected thereamong, and using any defects belonging to the remaining groups are used to calculate the classification criteria.
- classification is applied to defects belonging to the group selected in Step 503 .
- Such a process is executed for N times while changing the group to be selected in Steps 503 and 504 , thereby deriving the classification result for every teaching defect.
- Step 505 classification is applied to defects known with their true classification results, excepting for the teaching defects.
- the classification results derived as such in Steps 504 and 505 are compared with the true classification result.
- Step 506 comparison is made between the classification performances before and after deleting the teaching defect designated in Step 501 .
- Step 506 if determined as the performance being improved by deletion as such from the teaching defects, the corresponding defect will be a candidate as the one better to be deleted from the teaching defects.
- the classification performance there are a method for comparing the accuracy ratio among the classification classes, a method for comparison in terms of comprehensive performance, and a method of combination thereof. If the performance is improved as a result of both comparisons, the corresponding defects are regarded as the ones that have to be deleted. If the performance is reduced as a result of both comparisons, the defects are regarded as the ones essential for the teaching defects. And if the performance is improved as a result of either one comparison, the defects may be deleted or left as they are.
- Steps 501 to 506 The sequence from Steps 501 to 506 is executed while changing the teaching defect to be selected in Step 501 . In such a manner, unnecessary teaching data is extracted.
- Step 507 the teaching defects are ranked in order of showing the better performance improvement through comparison in Step 506 .
- the teaching defect is deleted one by one from the teaching defect cluster from the higher order, and in Step 509 , the classification criteria is set from the remaining teaching defect cluster.
- Step 510 is a step of performing classification using the classification criteria set in Step 509 .
- the process of Steps 502 to 505 is internally executed.
- Step 511 the classification result and the true classification are compared with each other, and when the performance is determined as being improved, the procedure returns to Step 508 for its process.
- the teaching defect lastly deleted from the teaching defect cluster in Step 512 is put back to the teaching defects again. Then the procedure returns to Step 508 to evaluate the next defect.
- Step 401 or 403 in the sequence of FIG. 4 incorporating the sequence of FIG. 5 is considered important for setting the best classification criteria.
- the sequence of FIG. 5 is a scheme guaranteeing to improve the comprehensive performance.
- the defects regarded as desirable to be deleted from the teaching defect cluster may be explicitly displayed for the user to allow him or her to check, and to perform performance comparison between a case where deletion is accordingly made and a case where no deletion is made.
- the method for classification criteria setting is found in JP-A-2001-256480 using a classification device of a teaching type. If the classification device is structured by a classification unit of a first layer and that of a layer lower thereto as the classification device described in JP-A-2001-135692, a type including semiconductor expertise by a rule logic, conventionally required is parameter setting using a manual to set classification criteria for the first layer. This is because even if teaching is made to the lower-layer classification unit for classification result, the lower-layer classification unit is not necessarily corresponding 1:1 to the first-layer classification unit. Thus, no such classification result can be obtained as expected by the discriminant used by the rule logic for the first layer.
- FIG. 6 shows such an exemplary case.
- a reference numeral 601 denotes a Graphical User Interface (GUI) for classification criteria setting, and a reference numeral 602 denotes a tag indicating the defect class of the lower layer.
- a reference numeral 603 denotes a tag indicating the defect class of the intermediate layer.
- GUI Graphical User Interface
- the tag 602 corresponding to CB(i) is linked to the tag 603 corresponding to CM(j) on the GUI.
- a thumb nail image 605 of the defect is displayed.
- the GUI 601 allows a user to input any instruction using a mouse pointer.
- the area 604 responsively displays the thumb hail image 605 of the defect belonging to the defect class indicated by the tag 602 .
- the thumb nail image 605 displayed in the area 604 is arbitrarily moved to one tag 603 via dragging and dropping popularly done in the GUI. In this manner, the true classification result for the lower-layer classification and the true classification result for the intermediate-layer classification can be simultaneously designated. By applying such an operation to every defect, all of the defects are provided with the classifying results for the intermediate layer and the lower layer.
- the user may set the classification criteria only for the lower layer.
- the conventionally-known automatic classification method known as multi-layer structure is a neutral network of back propagation type, for example.
- the lower layer and the intermediate layer can be set by classification criteria without answer-teaching to the intermediate layer. This method, however, requires a number of data, otherwise no appropriate classification criteria can be set.
- the requirement for the defect to be classified into the class i is to be classified into the intermediate class J, with which f2(i,J)(Fb)>f2(k,j)(Fb) is established with respect to an arbitrary class k.
- the cost evaluation expression is evaluated.
- the discriminant of the intermediate layer f1(j)(Fm) is set. In such a manner, the classification criteria of the intermediate layer is changed.
- f1(j)(Fm) is, as more specific example, is expressed by the following equation 2.
- a reference numeral 701 denotes an electron beam source
- a reference numeral 702 denotes a condenser lens
- Reference numerals 703 and 704 are deflection means in X and Y directions, respectively, for controlling the exposure positions of electron beams in the XY direction.
- the electron beams deflected by the deflection means are converged by an objective lens 705 , and irradiated onto a semiconductor wafer 707 being an observation object fixed onto an XY stage structured by 707 and 708 .
- Reference numerals 710 and 711 are both a reflected electron detector for detecting any reflected electrons coming from varying directions.
- a reference numeral 716 denotes an imaging unit for the detected electrons.
- Reference numerals 712 , 713 , and 714 all denote an A/D converter, which converts electronic signals detected by 709 , 710 , and 711 into digital data for storage into memory 715 .
- a reference numeral 717 is a defect position specification unit, which calculates any defect region from the images.
- a reference numeral 718 denotes a defect feature extraction unit, which extracts the defect image feature amount from the secondary defect images formed by the image memory 715 .
- the shape of the defect region or the brightness in the defect region calculated by 717 are regarded as defect feature amounts.
- the image feature amount extracted by 717 and the secondary defect image stored in 716 can be stored in secondary storage means 723 .
- a reference numeral 720 denotes classification criteria setting means, which realizes the classification criteria setting method already described in the above.
- a reference numeral 719 denotes classification means, which goes through defect classification using the classification criteria set by the classification criteria setting means 720 .
- the classification means 719 can classify in real time the image features derived by the image processing means 717 , and the defects stored in the secondary storage means 723 .
- a reference numeral 722 denotes a graphic terminal, which displays users with the images stored in 716 , the classification result derived by 721 , or the classification criteria set by 720 .
- This terminal can display also the GUI described by referring to FIG. 6 , whereby the users can show the classification result by using a manual. What is more, this terminal can display the relationship between the teaching defect number and the classification accuracy ratio described by referring to FIG. 3 .
- Described herein is an exemplary incorporation into the review SEM.
- the method of the present invention is surely incorporated in an optical review device, or used as a method for automatic defect classification unit incorporated on an inspection device.
- the appropriateness of automatic defect classification can be estimated, and classification criteria can be set with a guarantee for the better classification performance. Further, a resetting technique for the classification criteria with which the better classification performance is guaranteed in its entirety can be realized, having great significance as such.
Abstract
Description
f(i)(Fm,Fb)=max j(f1(j)(Fm)f2(i,j)(Fb)) Equation 1
f1(j)(Fm)=(Fm1>th1∩Fm2>th2)∪Fm3>th3 Equation 2
Although the accuracy ratio does not necessarily become maximum through such a change, the logical equation can be changed in such a manner as to make the accuracy ratio at least better than that of the initial state.
Claims (4)
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2003066581 | 2003-03-12 | ||
JP2003-066581 | 2003-03-12 |
Publications (2)
Publication Number | Publication Date |
---|---|
US20040234120A1 US20040234120A1 (en) | 2004-11-25 |
US7756320B2 true US7756320B2 (en) | 2010-07-13 |
Family
ID=33447014
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/794,267 Active 2026-06-26 US7756320B2 (en) | 2003-03-12 | 2004-03-04 | Defect classification using a logical equation for high stage classification |
Country Status (1)
Country | Link |
---|---|
US (1) | US7756320B2 (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110224932A1 (en) * | 2009-07-01 | 2011-09-15 | Kla-Tencor Corporation | Monitoring of time-varying defect classification performance |
US20130279796A1 (en) * | 2012-04-20 | 2013-10-24 | Applied Materials Israel Ltd. | Classifier readiness and maintenance in automatic defect classification |
US20140185918A1 (en) * | 2011-09-29 | 2014-07-03 | Hitachi High-Technologies Corporation | Charged particle beam apparatus that performs image classification assistance |
US9715723B2 (en) | 2012-04-19 | 2017-07-25 | Applied Materials Israel Ltd | Optimization of unknown defect rejection for automatic defect classification |
US10043264B2 (en) | 2012-04-19 | 2018-08-07 | Applied Materials Israel Ltd. | Integration of automatic and manual defect classification |
US10114368B2 (en) | 2013-07-22 | 2018-10-30 | Applied Materials Israel Ltd. | Closed-loop automatic defect inspection and classification |
Families Citing this family (41)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7756320B2 (en) * | 2003-03-12 | 2010-07-13 | Hitachi High-Technologies Corporation | Defect classification using a logical equation for high stage classification |
JP4681356B2 (en) * | 2005-06-13 | 2011-05-11 | 株式会社日立ハイテクノロジーズ | Defect automatic observation classification system, apparatus selection method, program, and observation apparatus in defect automatic observation classification system |
US8041103B2 (en) | 2005-11-18 | 2011-10-18 | Kla-Tencor Technologies Corp. | Methods and systems for determining a position of inspection data in design data space |
US7676077B2 (en) | 2005-11-18 | 2010-03-09 | Kla-Tencor Technologies Corp. | Methods and systems for utilizing design data in combination with inspection data |
US7570796B2 (en) | 2005-11-18 | 2009-08-04 | Kla-Tencor Technologies Corp. | Methods and systems for utilizing design data in combination with inspection data |
JP4644613B2 (en) | 2006-02-27 | 2011-03-02 | 株式会社日立ハイテクノロジーズ | Defect observation method and apparatus |
WO2008086282A2 (en) | 2007-01-05 | 2008-07-17 | Kla-Tencor Corporation | Methods and systems for using electrical information for a device being fabricated on a wafer to perform one or more defect-related functions |
US8213704B2 (en) | 2007-05-09 | 2012-07-03 | Kla-Tencor Corp. | Methods and systems for detecting defects in a reticle design pattern |
JP5425779B2 (en) * | 2007-08-20 | 2014-02-26 | ケーエルエー−テンカー・コーポレーション | A computer-implemented method for determining whether an actual defect is a potential systematic defect or a potentially random defect |
US20090259220A1 (en) * | 2008-04-09 | 2009-10-15 | Angiodynamics, Inc. | Treatment Devices and Methods |
US8139844B2 (en) | 2008-04-14 | 2012-03-20 | Kla-Tencor Corp. | Methods and systems for determining a defect criticality index for defects on wafers |
US9659670B2 (en) | 2008-07-28 | 2017-05-23 | Kla-Tencor Corp. | Computer-implemented methods, computer-readable media, and systems for classifying defects detected in a memory device area on a wafer |
JP5255953B2 (en) * | 2008-08-28 | 2013-08-07 | 株式会社日立ハイテクノロジーズ | Defect inspection method and apparatus |
US8775101B2 (en) | 2009-02-13 | 2014-07-08 | Kla-Tencor Corp. | Detecting defects on a wafer |
US8204297B1 (en) | 2009-02-27 | 2012-06-19 | Kla-Tencor Corp. | Methods and systems for classifying defects detected on a reticle |
US8112241B2 (en) | 2009-03-13 | 2012-02-07 | Kla-Tencor Corp. | Methods and systems for generating an inspection process for a wafer |
JP5537282B2 (en) | 2009-09-28 | 2014-07-02 | 株式会社日立ハイテクノロジーズ | Defect inspection apparatus and defect inspection method |
WO2011161900A1 (en) | 2010-06-25 | 2011-12-29 | 日本電気通信システム株式会社 | Information categorization system |
US8781781B2 (en) | 2010-07-30 | 2014-07-15 | Kla-Tencor Corp. | Dynamic care areas |
US9170211B2 (en) | 2011-03-25 | 2015-10-27 | Kla-Tencor Corp. | Design-based inspection using repeating structures |
US9087367B2 (en) | 2011-09-13 | 2015-07-21 | Kla-Tencor Corp. | Determining design coordinates for wafer defects |
US8831334B2 (en) | 2012-01-20 | 2014-09-09 | Kla-Tencor Corp. | Segmentation for wafer inspection |
US8826200B2 (en) | 2012-05-25 | 2014-09-02 | Kla-Tencor Corp. | Alteration for wafer inspection |
US9189844B2 (en) | 2012-10-15 | 2015-11-17 | Kla-Tencor Corp. | Detecting defects on a wafer using defect-specific information |
US9053527B2 (en) | 2013-01-02 | 2015-06-09 | Kla-Tencor Corp. | Detecting defects on a wafer |
US9134254B2 (en) | 2013-01-07 | 2015-09-15 | Kla-Tencor Corp. | Determining a position of inspection system output in design data space |
US9311698B2 (en) | 2013-01-09 | 2016-04-12 | Kla-Tencor Corp. | Detecting defects on a wafer using template image matching |
WO2014149197A1 (en) | 2013-02-01 | 2014-09-25 | Kla-Tencor Corporation | Detecting defects on a wafer using defect-specific and multi-channel information |
US9865512B2 (en) | 2013-04-08 | 2018-01-09 | Kla-Tencor Corp. | Dynamic design attributes for wafer inspection |
US9310320B2 (en) | 2013-04-15 | 2016-04-12 | Kla-Tencor Corp. | Based sampling and binning for yield critical defects |
US9286675B1 (en) * | 2014-10-23 | 2016-03-15 | Applied Materials Israel Ltd. | Iterative defect filtering process |
US9898811B2 (en) * | 2015-05-08 | 2018-02-20 | Kla-Tencor Corporation | Method and system for defect classification |
US20170069075A1 (en) * | 2015-09-04 | 2017-03-09 | Canon Kabushiki Kaisha | Classifier generation apparatus, defective/non-defective determination method, and program |
US10210380B2 (en) | 2016-08-09 | 2019-02-19 | Daon Holdings Limited | Methods and systems for enhancing user liveness detection |
US11115408B2 (en) | 2016-08-09 | 2021-09-07 | Daon Holdings Limited | Methods and systems for determining user liveness and verifying user identities |
US10628661B2 (en) | 2016-08-09 | 2020-04-21 | Daon Holdings Limited | Methods and systems for determining user liveness and verifying user identities |
US10217009B2 (en) * | 2016-08-09 | 2019-02-26 | Daon Holdings Limited | Methods and systems for enhancing user liveness detection |
US11138507B2 (en) * | 2017-09-28 | 2021-10-05 | Applied Materials Israel Ltd. | System, method and computer program product for classifying a multiplicity of items |
US11321633B2 (en) * | 2018-12-20 | 2022-05-03 | Applied Materials Israel Ltd. | Method of classifying defects in a specimen semiconductor examination and system thereof |
JP7203678B2 (en) | 2019-04-19 | 2023-01-13 | 株式会社日立ハイテク | Defect observation device |
CN115808382B (en) * | 2023-02-02 | 2023-04-21 | 深圳裕典通微电子科技有限公司 | Piezoelectric film on-line detection method and system applied to pressure sensor |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5463773A (en) * | 1992-05-25 | 1995-10-31 | Fujitsu Limited | Building of a document classification tree by recursive optimization of keyword selection function |
US5917332A (en) * | 1996-05-09 | 1999-06-29 | Advanced Micro Devices, Inc. | Arrangement for improving defect scanner sensitivity and scanning defects on die of a semiconductor wafer |
US5982933A (en) * | 1996-01-12 | 1999-11-09 | Canon Kabushiki Kaisha | Information processing method, information processing apparatus, and storage medium |
US6148099A (en) * | 1997-07-03 | 2000-11-14 | Neopath, Inc. | Method and apparatus for incremental concurrent learning in automatic semiconductor wafer and liquid crystal display defect classification |
JP2001135692A (en) | 1999-06-15 | 2001-05-18 | Applied Materials Inc | Adaptive hybrid automatic defect classifying method |
JP2001256480A (en) | 2000-03-09 | 2001-09-21 | Hitachi Ltd | Automatic picture classifying method and its device |
US20010042705A1 (en) * | 2000-05-18 | 2001-11-22 | Ryou Nakagaki | Method for classifying defects and device for the same |
US6480621B1 (en) * | 1995-08-08 | 2002-11-12 | Apple Computer, Inc. | Statistical classifier with reduced weight memory requirements |
US6564198B1 (en) * | 2000-02-16 | 2003-05-13 | Hrl Laboratories, Llc | Fuzzy expert system for interpretable rule extraction from neural networks |
US20030164942A1 (en) * | 1999-12-07 | 2003-09-04 | Kunihiko Take | Semiconductor wafer examination system |
US20030182251A1 (en) * | 2002-03-22 | 2003-09-25 | Donglok Kim | Accelerated learning in machine vision using artificially implanted defects |
US20040234120A1 (en) * | 2003-03-12 | 2004-11-25 | Hitachi High-Technologies Corporation | Defect classification method |
-
2004
- 2004-03-04 US US10/794,267 patent/US7756320B2/en active Active
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5463773A (en) * | 1992-05-25 | 1995-10-31 | Fujitsu Limited | Building of a document classification tree by recursive optimization of keyword selection function |
US6480621B1 (en) * | 1995-08-08 | 2002-11-12 | Apple Computer, Inc. | Statistical classifier with reduced weight memory requirements |
US5982933A (en) * | 1996-01-12 | 1999-11-09 | Canon Kabushiki Kaisha | Information processing method, information processing apparatus, and storage medium |
US5917332A (en) * | 1996-05-09 | 1999-06-29 | Advanced Micro Devices, Inc. | Arrangement for improving defect scanner sensitivity and scanning defects on die of a semiconductor wafer |
US6148099A (en) * | 1997-07-03 | 2000-11-14 | Neopath, Inc. | Method and apparatus for incremental concurrent learning in automatic semiconductor wafer and liquid crystal display defect classification |
JP2001135692A (en) | 1999-06-15 | 2001-05-18 | Applied Materials Inc | Adaptive hybrid automatic defect classifying method |
US6922482B1 (en) * | 1999-06-15 | 2005-07-26 | Applied Materials, Inc. | Hybrid invariant adaptive automatic defect classification |
US20030164942A1 (en) * | 1999-12-07 | 2003-09-04 | Kunihiko Take | Semiconductor wafer examination system |
US6564198B1 (en) * | 2000-02-16 | 2003-05-13 | Hrl Laboratories, Llc | Fuzzy expert system for interpretable rule extraction from neural networks |
JP2001256480A (en) | 2000-03-09 | 2001-09-21 | Hitachi Ltd | Automatic picture classifying method and its device |
US20010042705A1 (en) * | 2000-05-18 | 2001-11-22 | Ryou Nakagaki | Method for classifying defects and device for the same |
US20030182251A1 (en) * | 2002-03-22 | 2003-09-25 | Donglok Kim | Accelerated learning in machine vision using artificially implanted defects |
US20040234120A1 (en) * | 2003-03-12 | 2004-11-25 | Hitachi High-Technologies Corporation | Defect classification method |
Non-Patent Citations (1)
Title |
---|
Translation of JP2001-256480 (Date: Sep. 2001; Author: Nakagaki, Akira; Country: Japan). * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110224932A1 (en) * | 2009-07-01 | 2011-09-15 | Kla-Tencor Corporation | Monitoring of time-varying defect classification performance |
US8537349B2 (en) * | 2009-07-01 | 2013-09-17 | Kla-Tencor Corporation | Monitoring of time-varying defect classification performance |
US20140185918A1 (en) * | 2011-09-29 | 2014-07-03 | Hitachi High-Technologies Corporation | Charged particle beam apparatus that performs image classification assistance |
US9280814B2 (en) * | 2011-09-29 | 2016-03-08 | Hitachi High-Technologies Corporation | Charged particle beam apparatus that performs image classification assistance |
US9715723B2 (en) | 2012-04-19 | 2017-07-25 | Applied Materials Israel Ltd | Optimization of unknown defect rejection for automatic defect classification |
US10043264B2 (en) | 2012-04-19 | 2018-08-07 | Applied Materials Israel Ltd. | Integration of automatic and manual defect classification |
US20130279796A1 (en) * | 2012-04-20 | 2013-10-24 | Applied Materials Israel Ltd. | Classifier readiness and maintenance in automatic defect classification |
US9607233B2 (en) * | 2012-04-20 | 2017-03-28 | Applied Materials Israel Ltd. | Classifier readiness and maintenance in automatic defect classification |
US10114368B2 (en) | 2013-07-22 | 2018-10-30 | Applied Materials Israel Ltd. | Closed-loop automatic defect inspection and classification |
US10901402B2 (en) | 2013-07-22 | 2021-01-26 | Applied Materials Israel, Ltd. | Closed-loop automatic defect inspection and classification |
Also Published As
Publication number | Publication date |
---|---|
US20040234120A1 (en) | 2004-11-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7756320B2 (en) | Defect classification using a logical equation for high stage classification | |
JP4443270B2 (en) | Defect classification method | |
US7113628B1 (en) | Defect image classifying method and apparatus and a semiconductor device manufacturing process based on the method and apparatus | |
US7405835B2 (en) | High-accuracy pattern shape evaluating method and apparatus | |
US7602962B2 (en) | Method of classifying defects using multiple inspection machines | |
JP5957378B2 (en) | Defect observation method and defect observation apparatus | |
US8582864B2 (en) | Fault inspection method | |
JP4644613B2 (en) | Defect observation method and apparatus | |
US7734082B2 (en) | Defect inspection method | |
US6870169B2 (en) | Method and apparatus for analyzing composition of defects | |
JP5948138B2 (en) | Defect analysis support device, program executed by defect analysis support device, and defect analysis system | |
US20130294680A1 (en) | Image classification method and image classification apparatus | |
US20140037188A1 (en) | Defect review method and apparatus | |
US20080298670A1 (en) | Method and its apparatus for reviewing defects | |
WO2013153891A1 (en) | Charged particle beam apparatus | |
US20040028276A1 (en) | Method and its apparatus for classifying defects | |
WO2010023791A1 (en) | Method and device for defect inspection | |
CN109583465A (en) | The method and its system classify to the defects of semiconductor sample | |
KR102530950B1 (en) | Classification of Defects in Semiconductor Specimens | |
JP2021027212A (en) | System for deriving electrical properties and non-transient computer-readable medium | |
JP5374225B2 (en) | Wafer inspection condition determination method, wafer inspection condition determination system, and wafer inspection system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: HITACHI HIGH-TECHNOLOGIES CORPORATION, JAPAN Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:HONDA, TOSHIFUMI;MIYAMOTO, ATSUSHI;OKUDA, HIROHITO;REEL/FRAME:015604/0228;SIGNING DATES FROM 20040204 TO 20040223 Owner name: HITACHI HIGH-TECHNOLOGIES CORPORATION,JAPAN Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:HONDA, TOSHIFUMI;MIYAMOTO, ATSUSHI;OKUDA, HIROHITO;SIGNING DATES FROM 20040204 TO 20040223;REEL/FRAME:015604/0228 Owner name: HITACHI HIGH-TECHNOLOGIES CORPORATION, JAPAN Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:HONDA, TOSHIFUMI;MIYAMOTO, ATSUSHI;OKUDA, HIROHITO;SIGNING DATES FROM 20040204 TO 20040223;REEL/FRAME:015604/0228 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
FEPP | Fee payment procedure |
Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
MAFP | Maintenance fee payment |
Free format text: PAYMENT OF MAINTENANCE FEE, 8TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1552) Year of fee payment: 8 |
|
AS | Assignment |
Owner name: HITACHI HIGH-TECH CORPORATION, JAPAN Free format text: CHANGE OF NAME AND ADDRESS;ASSIGNOR:HITACHI HIGH-TECHNOLOGIES CORPORATION;REEL/FRAME:052259/0227 Effective date: 20200212 |
|
MAFP | Maintenance fee payment |
Free format text: PAYMENT OF MAINTENANCE FEE, 12TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1553); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY Year of fee payment: 12 |